Yu Laboratory


Home Bio-monitoring Mobile Robot

Home bio-monitoring mobile robot

Problem declaration TThe need of home care is increasing. Care givers want to grasp the activities in the house of elderly living alone for proper care plan planning.
Goal Acquisition of life rhythm and grasp of emergency situation
Identify human behavior continuously and in real time
Progress The robot can identify 6 motion (walking, standing, sitting, lying , bending, falling) based on the silhouette image while it follows the target using the RGB - D camera.

Grasping detection for ADL monitoring of elderly people at home using depth sensors

Problem declaration There is no one who always watches over the elderly living alone, so it is difficult to make an appropriate care plan for that person.
Goal To build a care plan that is suitable for living elderly alone, make a system that automatically monitors elderly living alone.
Among them, "motion of using tools" that often found in daily living movements is important in making a care plan, so focus on the motion of grasping the tools and create a system that automatically identifies the grasping motion with the camera
Progress Images are taken using a depth sensor, and grasping identification is performed by machine learning.

On the control of a home care support robot for CPR (CardioPulmonary Resuscitation )

Problem declaration Due to an increase in demand for home medical care, problems in home care will also increase.
One of them is a heart stop.
Goal It enables chest compressions which is one of cardiopulmonary resuscitation methods through cooperative work of people and home support robots with telecare method.
進捗 We completed a series of movements of searching, discovering, moving, and compressing chest compressions using chest compressions using a ready-made robot.

Healthcare services on the open networks (ex.)

Problem declaration In the healthcare service under the open network, the reliability of the sensing devices and the acquired data by them is not guaranteed.
Goal Developing the “Virtual sensor filter” which can gather the guaranteed data to be available for the services / applications.
Progress We optimized the sensor-embedded algorithm to detect Pulse to Pulse intervals (PPIs) which quality is available for heart rate variability analysis.

Intention estimation for people with dementia

Problem declaration People with dementia have difficulties to plan and complete ADLs.
Goal The system can estimate of intention of dementia. If the system detect a pause of behavior, it will suggest the next step for independent living. The system also give caregivers information about which observation are related to intention.
Progress We are considering dementia patients model about urination with probabilistic models.